Phase 1 — Meta initialization

When: first phase, or waypoints/meta.json doesn't exist.

Objective

Resolve indication identity and gather basic context. This is the foundation for all subsequent phases.

Actions

Step 1: Setup

mkdir -p "<workdir>/waypoints"

Step 2: Resolve indication identity - WebSearch for the indication name to confirm: - Standard clinical definition (if one exists) - ICD-10 codes (if applicable) - Common aliases and alternative names - Whether this is a recognized diagnostic entity or a biological state/concept - Note: some indications are not standard diagnoses (e.g., "immunosenescence", "ageing", "GLP-1 induced sarcopenia") — document this explicitly

Step 3: Establish hierarchy - Identify the parent indication from the input (additional_context may contain this) - Understand where this fits in the clinical taxonomy (therapy area, condition class) - Note cross-cutting classifications if applicable

Step 4: Quick trial landscape scan - Use clinical-trials MCP to search for trials: search_trials(condition="[indication name]") - Record total count, phase distribution, and status distribution - This gives a rough sense of how mature the indication is clinically

Step 5: Write outputs - Write waypoints/meta.json with indication identity - Initialize waypoints/sources_evaluated.json

Guidance

• This phase should complete in a single iteration
• If the indication has no ICD codes or standard definition, note this clearly — it affects every downstream section (epidemiology data will be scarce, regulatory path will be novel)
• Output references: see references/waypoint-schemas.md for meta.json format